OptiX and underlying Cuda blocks/threads

Can I somehow change (or at least check out) the amount of underlying cuda threads and blocks used in the optix-based application. I need this to compare performance of existing OptiX app in different parallelism scenarios, to measure (and to demonstrate it accordingly) the gain in speed of adding more cuda threads.
Can anyone help me with this?

Actually i have the same question, can we have a size control of block and grid in optix?

This troubles me a lot

Or could anyone just tell me it’s impossible in optix?